H<sub>∞</sub> optimal control technique is seen as a promising robust control technique that can effectively deal with the problems of model uncertainties. However, for H<sup>∞</sup> optimal control design to be successful one must be able to choose adequate performance and uncertainty weights. Until now, there is no a systematic way of choosing these weighting functions; they are generally selected based on trial and error. This approach not only is ineffective but also time consuming. In this paper, a systematic way of selecting the weighting functions in H<sub>∞</sub> optimal control is proposed. The selection of adequate weighting function is formulated as an optimization problem and solved using Population Based Incremental Learning (PBIL) Algorithm.